An efficient implementation of an optimal time domain ECG coder

This paper presents a practical implementation of an optimal time domain electrocardiogram (ECG) compression scheme based on graph theory. For a given set of signal samples, the optimal scheme produces the minimum set of samples required to satisfy a preset bound on the RMS error of the reconstructed signal. The problem is solved by a cubic dynamic programming algorithm, and we need to divide the signal into appropriate segments and process them independently. Both the size of the segments and how the signal is split up is of vital importance, and will affect the execution time and the coding performance. In this paper we present a windowing technique that take these effects into consideration. We will then illustrate how the block size affects the coding performance. Finally we will present coding results that show the superior performance of our approach compared to the well known FAN algorithm